Battle of S/PDIF vs USB: which is better?

[cheapsoundguy]

@amirm

Would you be willing to redo this test with the most modern equipment available?

E.g. a XMOS 208 or XMOS 216 USB based USB DAC vs a PCIE card with SPDIF out? (I think Creative still makes such a card).

Edit: It appears the last sound cards with Coax output have been discontinued. So this is basically an academic question at this point...

 

[Sal1950]

Personally I'd love to see an upgraded optical system, I aways loved Toslink for it's total isolation from ground issues, radiated noise and such. A newer system that would handle the highest rates for PCM and DSD is needed now for complete compatibliy with today sources.. My PC > Emotiva DC1 will handle 192 reliably as long as I keep the cable short (no more than about 1 meter).

 

[Trell]

@amirm

Would you be willing to redo this test with the most modern equipment available?

E.g. a XMOS 208 or XMOS 216 USB based USB DAC vs a PCIE card with SPDIF out? (I think Creative still makes such a card).

Edit: It appears the last sound cards with Coax output have been discontinued. So this is basically an academic question at this point...

You can find such PCIE cards in the pro audio segment. RME have such cards, has many features you might not need, and they are expensive.

I don’t think it will change amirm conclusion that much, if at all.

 

[Kal Rubinson]

Personally I'd love to see an upgraded optical system, I aways loved Toslink for it's total isolation from ground issues, radiated noise and such.

Are you aware of the Corning Optical cables? Aside from being inconveniently too long.........

USB 3.Optical™ Cables | OEM Optical Communication Solutions | Corning

USB 3.Optical™ Cables by Corning connects computers and devices at incredible speed and over longer distances. They’re thin, light and remarkably tough — Optical Cables by Corning can be bent, squeezed, and tangled.

www.corning.com

 

[Blumlein 88]

The original idea for USB 3.0 was to make the data carrying portion fiber optic on all the cables. I think a combination of expense, and the fact such cables are not conducive to sharp bending made them drop it as they developed wired methods to get the same and eventually even fast data transfer.

 

[Trell]

The original idea for USB 3.0 was to make the data carrying portion fiber optic on all the cables. I think a combination of expense, and the fact such cables are not conducive to sharp bending made them drop it as they developed wired methods to get the same and eventually even fast data transfer.

And USB 2 works very well for audio and can have longer cables with no issue.

 

[wwenze]

How about USB to SPDIF converter and get the worst of both worlds?

 

[Trell]

How about USB to SPDIF converter and get the worst of both worlds?

That is not a given that it will not work well . In fact this is how I’ve connected my RME UCX II audio interface to me RME ADI-2 DAC when I use it (ADAT instead of TOSLINK). I play audio to the UCX via USB and the DAC does it’s job. When I’m not using the UCX I play directly to the ADI-2.

 

[Blumlein 88]

How about USB to SPDIF converter and get the worst of both worlds?

No reason for this to automatically be so. I've a couple that work just fine. I forget if it is this thread or another where I show the jitter FFT's of Toslink and SPDIF from a USB to digital converter which has both outputs. They give identical results at the analog output of the DAC. Meaning the method of connection has no effect on the result.

 

[Overseas]

In lay terms and considering the topic, what about streaming by phone via USB OTG?
1. Anathema
2. Very good
3. Mehhh, not the best

 

[Sernyl]

On a PC MB that has both USB and optical outputs, I'd like to know the differences in their respective paths.
Which one is the simplest to feed my RME ADI-Pro ?

 

[Sal1950]

On a PC MB that has both USB and optical outputs, I'd like to know the differences in their respective paths.
Which one is the simplest to feed my RME ADI-Pro ?

Sound quality wise it really shouldn't matter. One advantage of Optical is it's total ground isolation and eliminates any possibility of ground loop hum, I use optical for my 2 channel DAC here. USB has many advantages for multich use but not a concern for you.
Simplest, One cable to plug in either way?

 

[Sernyl]

Thanks for the reply.
Yes, old-style stereo.
I just wondered which path of the signal (in the MB) had more complicated steps, i.e. more prone to disturbances.
I had tested both, with my ageing ears, and could not hear any diference.
Also tested an external XMOS USB-to-Toslink : idem.
As the ADI is distant from the PC, using 15 meters of priceless SPDIF plastic fibre (from Thomann) is more convenient.

 

[phofman]

I just wondered which path of the signal (in the MB) had more complicated steps, i.e. more prone to disturbances.

Technically:

MB TOSLINK: OS HDA driver -> internal bus HDA controller -> HDA bus -> HDA codec -> TOSLINK output -> TOSLINK receiver -> SPDIF receiver with master clock recovery -> I2S -> DAC chip with no external control (TOSLINK transfers only audio data)

USB async: OS USB-audio driver -> OS USB core driver merging requests for all devices hooked to the USB controller -> internal bus USB controller -> USB port -> USB receiver (e.g. XMOS) with dedicated audio clock, no clock recovery, transmitter is paced by the receiver -> I2S -> DAC chip, possibly with I2C control from the USB receiver (USB can transfer control signals)

As of reliability - the USB path is more CPU intensive as the software must assemble complete USB packets. But no issue for modern hardware.

As of clock quality - the USB path allows master clock by the DAC (built-in pace feedback), SPDIF/TOSLINK requires clock recovery at the SPDIF receiver (the transmitter is always master, no clock feedback)

As of noise - TOSLINK is galvanically isolated, USB shares common ground (noisy ground loop currents can occur), easily fixed by an inexpensive 480Mbps USB isolator isolator.

Practically - no audible difference, unless USB ground loop issues arise.

 

[Sernyl]

Technically:

MB TOSLINK: OS HDA driver -> internal bus HDA controller -> HDA bus -> HDA codec -> TOSLINK output -> TOSLINK receiver -> SPDIF receiver with master clock recovery -> I2S -> DAC chip with no external control (TOSLINK transfers only audio data)

USB async: OS USB-audio driver -> OS USB core driver merging requests for all devices hooked to the USB controller -> internal bus USB controller -> USB port -> USB receiver (e.g. XMOS) with dedicated audio clock, no clock recovery, transmitter is paced by the receiver -> I2S -> DAC chip, possibly with I2C control from the USB receiver (USB can transfer control signals)

As of reliability - the USB path is more CPU intensive as the software must assemble complete USB packets. But no issue for modern hardware.

As of clock quality - the USB path allows master clock by the DAC (built-in pace feedback), SPDIF/TOSLINK requires clock recovery at the SPDIF receiver (the transmitter is always master, no clock feedback)

As of noise - TOSLINK is galvanically isolated, USB shares common ground (noisy ground loop currents can occur), easily fixed by an inexpensive 480Mbps USB isolator isolator.

Practically - no audible difference, unless USB ground loop issues arise.

Thanks a lot !
Anyway, a 15 meters USB cable would have its owm problems...

 

[phofman]

Yes, 15m would require an active USB cable with repeater, probably better to use a good quality toslink.